Skip to content

Pull requests: sgl-project/sglang

New pull request New
793 Open 7,808 Closed
793 Open 7,808 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

Super tiny fix naming in bench serving scripts run-ci
#12515 opened Nov 2, 2025 by fzyzcjy Loading…
4 tasks
2
Supports direct retrieval of tokenizer content and chat template content from the worker. run-ci
#12513 opened Nov 2, 2025 by whybeyoung Loading…
7
Reduce time it takes to import SGLang
#12510 opened Nov 2, 2025 by raayandhar • Draft
3 of 4 tasks
2
[GDN] Fuse b.sigmoid(), fused_gdn_gating and unsqueeze into one kernel: up to 0.85% e2e speedup run-ci
#12508 opened Nov 2, 2025 by byjiang1996 Loading…
4 tasks done
3
Add prefix for torch symm mem run-ci
#12506 opened Nov 2, 2025 by yuan-luo Loading…
4 tasks
3
To implement attention-sink
#12504 opened Nov 2, 2025 by haotang1995 Loading…
4
Fuse MLA RoPE + FP8 Quantization + KV Cache Write into Single CUDA Kernel high priority
#12503 opened Nov 2, 2025 by YAMY1234 • Draft
4 tasks
@Qiaolin-Yu
4
[Refact] Remove hardcoded KV cache dimension in MLATokenToKVPool run-ci
#12502 opened Nov 1, 2025 by Johnsonms Loading…
4 tasks done
4
fix: Missing @dataclass for ExpertDistributionReq
#12501 opened Nov 1, 2025 by SecretSettler Loading…
4 tasks done
2
fix: relax padding for fp4 in weight loading run-ci
#12497 opened Nov 1, 2025 by b8zhong Loading…
1
4
WIP: [Bug] fix blocking operations in coroutines
#12495 opened Nov 1, 2025 by howardlau1999 Loading…
4 tasks
11
fix: Excessive preemption occurs when preempting running requests to schedule new prefill requests.
#12494 opened Nov 1, 2025 by CLFutureX Loading…
3
[Feature] Ascend support enable-mixed-chunk version2
#12491 opened Nov 1, 2025 by MichelleWu351 • Draft
6
[Feature] Ascend support enable-mixed-chunk
#12490 opened Nov 1, 2025 by MichelleWu351 • Draft
5
fix typo of args description in sglang.profiler
#12486 opened Nov 1, 2025 by ai-easy-cpu Loading…
4 tasks
3
WIP: diffusion run-ci
#12484 opened Nov 1, 2025 by mickqian Loading…
4 tasks
9
[ServerArgs] allow --mamba-ssm-dtype extend run-ci
#12481 opened Nov 1, 2025 by hanming-lu Loading…
4 tasks
3
Add the Nvidia-ModelOPT FP8 quantization test case run-ci
#12478 opened Oct 31, 2025 by jingyu-ml Loading…
2 of 4 tasks
@Edwardf0t1 @jingyu-ml
5
[router] add WASM support for middleware router run-ci
#12471 opened Oct 31, 2025 by tonyluj Loading…
1 of 4 tasks
16
test: support return logprobs in bench_offline_throughput test
#12462 opened Oct 31, 2025 by aftersnow Loading…
4 tasks
3
Logging cache_hit_rate only for prefill and speculative_decoding metrics only for decode
#12460 opened Oct 31, 2025 by MMuzzammil1 Loading…
1 of 4 tasks
1
3
fix: use model-specific params to resize images for qwen-vl series models run-ci
#12458 opened Oct 31, 2025 by yangsijia-serena Loading…
4 tasks
1
3
[fix] Handle escaped characters in GLM tool call parser to prevent double serialization run-ci
#12456 opened Oct 31, 2025 by soaringk Loading…
2 of 4 tasks
3
[Grammar Fix] GLM-4-MOE self.first_k_dense_replace is undefined.
#12455 opened Oct 31, 2025 by zRzRzRzRzRzRzR Loading…
@ch-wan
2
[Fix] concat_mla_absorb_q_kernel fails for long inputs
#12453 opened Oct 31, 2025 by bingps Loading…
2 of 4 tasks
6
ProTip! What’s not been updated in a month: updated:<2025-10-02.